
import pandas as pd
import numpy as np
import plotly.express as px
import seaborn as sns
import matplotlib.pyplot as plt
from sklearn.cluster import KMeans

from sklearn.neighbors import NearestNeighbors

from sklearn.preprocessing import StandardScaler


from sklearn.metrics import silhouette_score
from sklearn.metrics import adjusted_rand_score

from scipy import stats
from kmodes.kmodes import KModes
student_no = 3914796


#read in the data and create a dataframe
file = '/Users/gretastojanovic/Desktop/DS/Python/RMIT_Python/Assignment_2/A2_HR_Employee_Data.csv'
raw = pd.read_csv(file)
df = pd.DataFrame(raw)
df.info() #get overview of the data

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 1470 entries, 0 to 1469
Data columns (total 22 columns):
 #   Column                    Non-Null Count  Dtype  
---  ------                    --------------  -----  
 0   EmployeeID                1470 non-null   int64  
 1   Age                       1470 non-null   int64  
 2   Resigned                  1470 non-null   object 
 3   BusinessTravel            1470 non-null   object 
 4   BusinessUnit              1470 non-null   object 
 5   EducationLevel            1470 non-null   int64  
 6   Gender                    1470 non-null   object 
 7   JobSatisfaction           1470 non-null   int64  
 8   MaritalStatus             1470 non-null   object 
 9   MonthlyIncome             1470 non-null   int64  
 10  NumCompaniesWorked        1470 non-null   int64  
 11  OverTime                  1470 non-null   object 
 12  PercentSalaryHike         1470 non-null   int64  
 13  PerformanceRating         1470 non-null   int64  
 14  AverageWeeklyHoursWorked  1470 non-null   float64
 15  TotalWorkingYears         1470 non-null   int64  
 16  TrainingTimesLastYear     1470 non-null   int64  
 17  WorkLifeBalance           1470 non-null   int64  
 18  YearsAtCompany            1470 non-null   int64  
 19  YearsInRole               1470 non-null   int64  
 20  YearsSinceLastPromotion   1470 non-null   int64  
 21  YearsWithCurrManager      1470 non-null   int64  
dtypes: float64(1), int64(15), object(6)
memory usage: 252.8+ KB


#plot frequency of all variables
sns.set(style="dark")  
fig,axs = plt.subplots(7,3, figsize = (15,12))
sns.histplot(data=df, x="EmployeeID", kde=True, color="skyblue", ax=axs[0, 0])

sns.histplot(data=df, x="Age", kde=True, color="skyblue", ax=axs[0, 0])
sns.histplot(data=df, x="EducationLevel", kde=True, color="olive", ax=axs[0, 1])
sns.histplot(data=df, x="JobSatisfaction", kde=True, color="gold", ax=axs[0, 2])
sns.histplot(data=df, x="MonthlyIncome", kde=True, color="teal", ax=axs[1, 0])
sns.histplot(data=df, x="NumCompaniesWorked", kde=True, color="teal", ax=axs[1, 1])
sns.histplot(data=df, x="PercentSalaryHike", kde=True, color="blue", ax=axs[1, 2])
sns.histplot(data=df, x="PerformanceRating", kde=True, color="purple", ax=axs[2,0])
sns.histplot(data=df, x="AverageWeeklyHoursWorked", kde=True, color="skyblue", ax=axs[2, 1])
sns.histplot(data=df, x="TotalWorkingYears", kde=True, color="olive", ax=axs[2, 2])
sns.histplot(data=df, x="TrainingTimesLastYear", kde=True, color="gold", ax=axs[3, 0])
sns.histplot(data=df, x="WorkLifeBalance", kde=True, color="teal", ax=axs[3, 1])
sns.histplot(data=df, x="YearsAtCompany", kde=True, color="teal", ax=axs[3,2])
sns.histplot(data=df, x="YearsInRole", kde=True, color="blue", ax=axs[4, 0])
sns.histplot(data=df, x="YearsSinceLastPromotion", kde=True, color="purple", ax=axs[4,1])
sns.histplot(data=df, x="YearsWithCurrManager", kde=True, color="skyblue", ax=axs[4, 2])
sns.histplot(data=df, x="Resigned", kde=True, color="skyblue", ax=axs[5, 0])
sns.histplot(data=df, x="BusinessTravel", kde=True, color="olive", ax=axs[5, 1])
sns.histplot(data=df, x="BusinessUnit", kde=True, color="gold", ax=axs[5, 2])
sns.histplot(data=df, x="Gender", kde=True, color="teal", ax=axs[6, 0])
sns.histplot(data=df, x="MaritalStatus", kde=True, color="blue", ax=axs[6, 1])
sns.histplot(data=df, x="OverTime", kde=True, color="purple", ax=axs[6,2])

fig.tight_layout()


df.describe()


#get numeric columns
numerics = list(df.select_dtypes(['int64', 'float64']).columns)
# create numeric dataframe 
numerics = df[numerics]
numerics = pd.DataFrame(numerics)
#remove outliers    
#numerics = numerics[(np.abs(stats.zscore(numerics)) < 3).all(axis=1)]
Q1 = numerics.quantile(0.25)
Q3 = numerics.quantile(0.75)
IQR = Q3 - Q1
numerics = numerics[~((numerics < (Q1 - 1.5 * IQR)) |(numerics > (Q3 + 1.5 * IQR))).any(axis=1)]

#get object columns
objectcols = list(df.select_dtypes(['object']).columns)
objectcols.append('EmployeeID')
#create dataframe from object columns
objectdf = df[objectcols]
objectdf = pd.DataFrame(objectdf)
newdf = objectdf.merge(numerics, how='inner', on='EmployeeID')
onehot = pd.get_dummies(newdf)


#plot correlation matrix
plt.figure(figsize = (30,30))
corrmat = onehot.corr()
sns.heatmap(corrmat, annot = True, square = True, cmap = 'YlGnBu')
plt.show()


# show top 10 correlations with Resigned variable
corr = onehot.corr()
corr = pd.DataFrame(corr)
#corr['Resigned_Yes'] = abs(corr['Resigned_Yes'].values)
res_yes = pd.DataFrame(corr['Resigned_Yes'])
res_yes['abs'] = abs(res_yes['Resigned_Yes'].values)
res_yes.sort_values(by = 'abs', ascending = False).head(14)


res_yes_df = df[['Resigned', 'AverageWeeklyHoursWorked', 'OverTime', 'YearsAtCompany', 
                 'TotalWorkingYears', 'YearsInRole', 'Age', 'YearsWithCurrManager', 
                 'MaritalStatus', 'MonthlyIncome', 'BusinessTravel', 'BusinessUnit']] 

res_yes_df_onehot = onehot[['Resigned_Yes', 'AverageWeeklyHoursWorked', 'OverTime_Yes', 'YearsAtCompany', 
                 'TotalWorkingYears', 'YearsInRole', 'Age', 'YearsWithCurrManager', 
                 'MaritalStatus_Single', 'MonthlyIncome', 'BusinessTravel_Travel_Frequently', 'BusinessUnit_Consultants']]


res_yes_df_data = res_yes_df.drop(['Resigned'], axis =1)
res_yes_df_data = pd.DataFrame(res_yes_df_data)


# pairplot for numeric columns
sns.pairplot(res_yes_df, hue='Resigned') 
plt.show()


#pairplot for all columns via one hot encoding


sns.pairplot(res_yes_df_onehot, hue='Resigned_Yes') 
plt.show()


def plot_elbow_graph(dataset, k_range=range(1, 11)): 

    # A list holds the inertia values for each k 

    inertia_lst = [] 

    for k in k_range: 

        km_model = KMeans(n_clusters=k, random_state=student_no) 

        km_model.fit(dataset) 

        inertia_lst.append(km_model.inertia_) 

    plt.figure(figsize=(10,6)) 

    plt.plot(k_range, inertia_lst) 

    plt.xticks(k_range) 

    plt.xlabel("Number of Clusters") 

    plt.ylabel("Inertia") 

    plt.show()
    
def plot_silhouette_graph(dataset, k_range=range(2, 11)): 

    # A list holds the silhouette coefficients for each k 

    silhouette_coefficients = [] 

    for k in k_range: 

        km_model = KMeans(n_clusters=k, random_state=student_no) 

        km_model.fit(dataset) 

        score = silhouette_score(dataset, km_model.labels_) 

        silhouette_coefficients.append(score) 

    plt.figure(figsize=(10,6)) 

    plt.plot(k_range, silhouette_coefficients) 

    plt.xticks(k_range) 

    plt.xlabel("Number of Clusters") 

    plt.ylabel("Silhouette Coefficient") 

    plt.show()


#use the one hot encoded version of the data to obtain an elbow 
#graph result (data must be numeric)

plot_elbow_graph(res_yes_df_onehot, k_range=range(2, 20)) 

plot_silhouette_graph(res_yes_df_onehot, k_range=range(2, 20))


#plotting the Resigned variable as a histogram
fig = hist = px.histogram(df, x = "Resigned", 
                          labels = {"Resigned": "Resignation Status", 
                                    "count": "Number of Employees"}, 
            title = "237 of 1470 Employees have Resigned (16%)")

fig.show()


km = KModes(n_clusters=3, init='Huang', n_init=5, verbose=1)

clusters = km.fit_predict(res_yes_df_data)
clusters = km.fit_predict(res_yes_df_data)


km_df = res_yes_df.copy()

# add a new column to the df and assign it the clustes labels
km_df['cluster'] = clusters


# Print the cluster centroids
print(km.cluster_centroids_)
print(km_df.loc[:,['cluster','Resigned']].value_counts().sort_index())

Init: initializing centroids
Init: initializing clusters
Starting iterations...
Run 1, iteration: 1/100, moves: 624, cost: 8935.0
Run 1, iteration: 2/100, moves: 198, cost: 8914.0
Run 1, iteration: 3/100, moves: 9, cost: 8913.0
Run 1, iteration: 4/100, moves: 0, cost: 8913.0
Init: initializing centroids
Init: initializing clusters
Starting iterations...
Run 2, iteration: 1/100, moves: 374, cost: 9241.0
Run 2, iteration: 2/100, moves: 112, cost: 9226.0
Run 2, iteration: 3/100, moves: 142, cost: 9076.0
Run 2, iteration: 4/100, moves: 46, cost: 9076.0
Init: initializing centroids
Init: initializing clusters
Starting iterations...
Run 3, iteration: 1/100, moves: 223, cost: 9295.0
Run 3, iteration: 2/100, moves: 2, cost: 9295.0
Init: initializing centroids
Init: initializing clusters
Starting iterations...
Run 4, iteration: 1/100, moves: 312, cost: 8950.0
Run 4, iteration: 2/100, moves: 252, cost: 8950.0
Init: initializing centroids
Init: initializing clusters
Starting iterations...
Run 5, iteration: 1/100, moves: 227, cost: 9154.0
Run 5, iteration: 2/100, moves: 25, cost: 9154.0
Best run was number 1
Init: initializing centroids
Init: initializing clusters
Starting iterations...
Run 1, iteration: 1/100, moves: 428, cost: 8943.0
Run 1, iteration: 2/100, moves: 75, cost: 8942.0
Run 1, iteration: 3/100, moves: 0, cost: 8942.0
Init: initializing centroids
Init: initializing clusters
Starting iterations...
Run 2, iteration: 1/100, moves: 435, cost: 9195.0
Run 2, iteration: 2/100, moves: 411, cost: 8917.0
Run 2, iteration: 3/100, moves: 114, cost: 8917.0
Init: initializing centroids
Init: initializing clusters
Starting iterations...
Run 3, iteration: 1/100, moves: 573, cost: 8986.0
Run 3, iteration: 2/100, moves: 183, cost: 8986.0
Init: initializing centroids
Init: initializing clusters
Starting iterations...
Run 4, iteration: 1/100, moves: 324, cost: 9011.0
Init: initializing centroids
Init: initializing clusters
Starting iterations...
Run 5, iteration: 1/100, moves: 395, cost: 8946.0
Run 5, iteration: 2/100, moves: 37, cost: 8946.0
Best run was number 2
[['40.0' 'No' '5' '6' '2' '31' '2' 'Married' '2342' 'Travel_Rarely'
  'Consultants']
 ['40.0' 'No' '1' '1' '0' '35' '0' 'Married' '2559' 'Travel_Rarely'
  'Consultants']
 ['40.0' 'No' '10' '10' '7' '34' '7' 'Single' '5343' 'Travel_Rarely'
  'Sales']]
cluster  Resigned
0        No          775
         Yes         107
1        No          195
         Yes          80
2        No          263
         Yes          50
dtype: int64


km_df[km_df['cluster'] == 1].describe()


from kmodes.kprototypes import KPrototypes
#dataframe to an array
smart_array = res_yes_df.values
#converting numerical columns datatype as float
smart_array[:, 3] = smart_array[:,3].astype(float)
smart_array[:, 4] = smart_array[:,4].astype(float)
smart_array[:, 5] = smart_array[:,5].astype(float)
smart_array[:, 6] = smart_array[:,6].astype(float)
smart_array[:, 7] = smart_array[:,7].astype(float)
smart_array[:, 9] = smart_array[:,9].astype(float)

#index of categorical columns
categorical_index = [0] + [2] + [8] + [10, 11]
print('done')

done


# Function for plotting elbow curve
def plot_elbow_curve(start, end, data):
    no_of_clusters = list(range(start, end+1))
    cost_values = []

    for k in no_of_clusters:
        test_model = KPrototypes(n_clusters=k, init='Huang', 
                                 random_state=student_no)
        test_model.fit_predict(data, categorical=categorical_index)
        cost_values.append(test_model.cost_)

    sns.set_theme(style="whitegrid", palette="bright", font_scale=1.2)

    plt.figure(figsize=(15, 7))
    ax = sns.lineplot(x=no_of_clusters, y=cost_values, marker="o", 
                      dashes=False)
    ax.set_title('Elbow curve', fontsize=18)
    ax.set_xlabel('No of clusters', fontsize=14)
    ax.set_ylabel('Cost', fontsize=14)
    ax.set(xlim=(start-0.1, end+0.1))
    plt.plot()

# Plotting elbow curve for k=2 to k=10
plot_elbow_curve(2,10,smart_array)


model_6 = KPrototypes(n_clusters=6, init='Huang', random_state=student_no, 
                      n_jobs=-1)
model_6.fit_predict(smart_array, categorical=categorical_index)
print(model_6.cost_)
#new column for cluster labels associated with each subject
res_yes_df['labels6'] = model_6.labels_
res_yes_df.head()

748428046.2954373

/var/folders/bg/9bc3wg_14jl4dwfytmh7brfh0000gn/T/ipykernel_22464/2095390600.py:6: SettingWithCopyWarning:


A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy


res_yes_df.loc[:,['labels6','Resigned']].value_counts().sort_index().to_frame().rename({0:'count'}, axis=1)


res_yes_df1 = res_yes_df[res_yes_df['labels6'] == 1]
res_yes_df1.describe()


res_yes_df1.describe(include=np.object)

/var/folders/bg/9bc3wg_14jl4dwfytmh7brfh0000gn/T/ipykernel_22464/461098861.py:1: DeprecationWarning:

`np.object` is a deprecated alias for the builtin `object`. To silence this warning, use `object` by itself. Doing this will not modify any behavior and is safe. 
Deprecated in NumPy 1.20; for more details and guidance: https://numpy.org/devdocs/release/1.20.0-notes.html#deprecations


fig, axes = plt.subplots(nrows=1, ncols=3, figsize=(10,5), sharex=True, 
                         sharey=True)
# creating a scatterplot of the clustered data on the first sub-axes 
sns.scatterplot(data=res_yes_df1, x='AverageWeeklyHoursWorked', 
                y='MonthlyIncome',
                hue='Resigned', palette='colorblind', ax=axes[0], alpha=0.8)
# creating another scatter plot of the centroids on the first sub-axes
# replace hue=range() with the parameter k
sns.scatterplot(data=res_yes_df, x='AverageWeeklyHoursWorked', 
                y='MonthlyIncome',
                hue='Resigned', palette='colorblind', ax=axes[1],
                alpha=0.8)
# # setting the title for the first sub-axes
# # add the parameter k to the title
# axes[0].set_title(f'Clustered data with k={k}')

# creating a scatterplot of the true labeled data on the second sub-axes 
sns.scatterplot(data=res_yes_df, x='AverageWeeklyHoursWorked', 
                y='MonthlyIncome',
                hue='labels6', palette='colorblind', ax=axes[2]
                ,alpha=0.8)
# setting the title for the second sub-axes
axes[1].set_title('Original data')
plt.show()


fig, axes = plt.subplots(nrows=1, ncols=2, figsize=(10,5), sharex=True, 
                         sharey=True)
# creating a scatterplot of the clustered data on the first sub-axes 
sns.scatterplot(data=res_yes_df, x='AverageWeeklyHoursWorked', y='Age',
                hue='labels6', palette='colorblind', ax=axes[0], 
                alpha=0.8)


# creating a scatterplot of the true labeled data on the second sub-axes 
sns.scatterplot(data=res_yes_df, x='AverageWeeklyHoursWorked', 
                y='Age',
                hue='Resigned', palette='colorblind', ax=axes[1],
                alpha=0.8)
# setting the title for the second sub-axes
axes[1].set_title('Original data')
plt.show()


Remuneration = onehot[['OverTime_Yes', 'MonthlyIncome', 'Resigned_Yes', 
                       'Age']]
# standardizing the data
from sklearn.preprocessing import StandardScaler
scaler = StandardScaler()
Remuneration_scaled = scaler.fit_transform(Remuneration)


#find optimal number of clusters:
wcssa = []
for i in range(1,10):
    kmeans = KMeans(n_clusters = i, init = 'k-means++', random_state = 0)
    kmeans.fit(Remuneration_scaled)
    wcssa.append(kmeans.inertia_)
plt.plot(range(1,10),wcssa)
plt.show()


# k means using 6 clusters and k-means++ initialization
model = KMeans(n_clusters = 4, init='k-means++', random_state = student_no)
rres = model.fit(Remuneration_scaled)
Remuneration["cluster"] = rres.predict(Remuneration_scaled)
Remuneration["cluster"].value_counts()

/var/folders/bg/9bc3wg_14jl4dwfytmh7brfh0000gn/T/ipykernel_22464/2639930091.py:4: SettingWithCopyWarning:


A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead

See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy

1    352
2    175
0    155
3    138
Name: cluster, dtype: int64


#this is a scatterplot of the original dataframe Monthly 
#Income vs Age with the Resigned status highlighted in red

fig, ax = plt.subplots()

colors = {0.0:'cornflowerblue', 1.0: 'red'}

ax.scatter(onehot['Age'], onehot['MonthlyIncome'], 
           c=onehot['Resigned_Yes'].map(colors))
plt.show()


#the below plots is the result of K-means clustering analysis for 
#Monthly Income vs Age
#the black colour shows a pattern similar to that of the 'resigned' 
#variable indicating the model has identified a cluster of employees 
#that would be
#likely to resign based on Age and Income as inputs 

fig, ax = plt.subplots(1)

# plot clustering result
plt.scatter(Remuneration.iloc[:,3], 
            Remuneration.iloc[:,1],
            c=Remuneration.cluster, 
            alpha=0.8)

plt.show()


R0 = Remuneration[Remuneration['cluster'] == 0]
R0.describe()


R1 = Remuneration[Remuneration['cluster'] == 1]
R1.describe()


R3 = Remuneration[Remuneration['cluster'] == 3]
R3.describe()

	EmployeeID	Age	EducationLevel	JobSatisfaction	MonthlyIncome	NumCompaniesWorked	PercentSalaryHike	PerformanceRating	AverageWeeklyHoursWorked	TotalWorkingYears	TrainingTimesLastYear	WorkLifeBalance	YearsAtCompany	YearsInRole	YearsSinceLastPromotion	YearsWithCurrManager
count	1470.000000	1470.000000	1470.000000	1470.000000	1470.000000	1470.000000	1470.000000	1470.000000	1470.000000	1470.000000	1470.000000	1470.000000	1470.000000	1470.000000	1470.000000	1470.000000
mean	5491.933333	36.923810	2.912925	2.728571	6502.931293	2.693197	15.209524	3.153741	43.042177	11.279592	2.799320	2.761224	7.008163	4.229252	2.187755	4.123129
std	2573.640304	9.135373	1.024165	1.102846	4707.956783	2.498009	3.659938	0.360824	5.389864	7.780782	1.289271	0.706476	6.126525	3.623137	3.222430	3.568136
min	1006.000000	18.000000	1.000000	1.000000	1009.000000	0.000000	11.000000	3.000000	40.000000	0.000000	0.000000	1.000000	0.000000	0.000000	0.000000	0.000000
25%	3286.250000	30.000000	2.000000	2.000000	2911.000000	1.000000	12.000000	3.000000	40.000000	6.000000	2.000000	2.000000	3.000000	2.000000	0.000000	2.000000
50%	5484.500000	36.000000	3.000000	3.000000	4919.000000	2.000000	14.000000	3.000000	40.000000	10.000000	3.000000	3.000000	5.000000	3.000000	1.000000	3.000000
75%	7766.250000	43.000000	4.000000	4.000000	8379.000000	4.000000	18.000000	3.000000	47.000000	15.000000	3.000000	3.000000	9.000000	7.000000	3.000000	7.000000
max	9991.000000	60.000000	5.000000	4.000000	19999.000000	9.000000	25.000000	4.000000	71.000000	40.000000	6.000000	4.000000	40.000000	18.000000	15.000000	17.000000

	Resigned_Yes	abs
Resigned_Yes	1.000000	1.000000
Resigned_No	-1.000000	1.000000
AverageWeeklyHoursWorked	0.305732	0.305732
OverTime_No	-0.220164	0.220164
OverTime_Yes	0.220164	0.220164
YearsAtCompany	-0.191745	0.191745
TotalWorkingYears	-0.187664	0.187664
YearsInRole	-0.172654	0.172654
Age	-0.168777	0.168777
YearsWithCurrManager	-0.167105	0.167105
MaritalStatus_Single	0.163453	0.163453
MonthlyIncome	-0.151037	0.151037
BusinessTravel_Travel_Frequently	0.123242	0.123242
BusinessUnit_Consultants	-0.104012	0.104012

	AverageWeeklyHoursWorked	YearsAtCompany	TotalWorkingYears	YearsInRole	Age	YearsWithCurrManager	MonthlyIncome	cluster
count	275.000000	275.000000	275.000000	275.000000	275.000000	275.000000	275.000000	275.0
mean	43.680000	2.745455	8.643636	1.221818	35.254545	0.898182	5252.650909	1.0
std	5.853391	4.446019	8.031207	2.941938	9.870730	2.460750	4431.200704	0.0
min	40.000000	0.000000	0.000000	0.000000	18.000000	0.000000	1009.000000	1.0
25%	40.000000	1.000000	1.000000	0.000000	29.000000	0.000000	2389.000000	1.0
50%	40.000000	1.000000	7.000000	0.000000	35.000000	0.000000	3038.000000	1.0
75%	48.500000	1.000000	15.000000	0.000000	41.000000	0.000000	6645.000000	1.0
max	67.000000	33.000000	34.000000	15.000000	60.000000	15.000000	19613.000000	1.0

	AverageWeeklyHoursWorked	YearsAtCompany	TotalWorkingYears	YearsInRole	Age	YearsWithCurrManager	MonthlyIncome	labels6
count	489.000000	489.000000	489.000000	489.000000	489.000000	489.000000	489.000000	489.0
mean	43.306748	3.940695	6.042945	2.492843	32.691207	2.480573	2571.952965	1.0
std	5.809826	3.276619	4.437198	2.560027	8.714315	2.677343	522.653393	0.0
min	40.000000	0.000000	0.000000	0.000000	18.000000	0.000000	1009.000000	1.0
25%	40.000000	1.000000	3.000000	0.000000	27.000000	0.000000	2267.000000	1.0
50%	40.000000	3.000000	6.000000	2.000000	31.000000	2.000000	2559.000000	1.0
75%	48.000000	5.000000	8.000000	3.000000	37.000000	3.000000	2909.000000	1.0
max	70.000000	20.000000	20.000000	13.000000	59.000000	17.000000	3622.000000	1.0

	OverTime_Yes	MonthlyIncome	Resigned_Yes	Age	cluster
count	138.000000	138.000000	138.0	138.000000	138.0
mean	0.514493	3999.586957	1.0	31.804348	3.0
std	0.501611	2473.391917	0.0	8.414325	0.0
min	0.000000	1081.000000	1.0	18.000000	3.0
25%	0.000000	2336.750000	1.0	26.250000	3.0
50%	1.000000	2897.500000	1.0	30.500000	3.0
75%	1.000000	5287.500000	1.0	35.000000	3.0
max	1.000000	13610.000000	1.0	56.000000	3.0

Practical Data Science With Python, RMIT¶

Assessment 2 - Greta Stojanovic s3914796¶

Remove Outliers to Enable Better Clustering Outcome¶

Explore Correlations Between Resigned Variable and Other Features¶

Assign Top Correlated Variables to DataFrames for Processing¶

Plot Elbow Graph to Determine Number of Clusters¶

Model 1: Using KModes (this removes categoricals)¶

Cluster 1 has the highest ratio of resignations at 32%. 339 in the cluster. 1470 in entire dataset.¶

Model 2: Applying KPrototypes which Includes Categoricals in the Modelling¶

Cluster 1 from the KPrototypes analysis has the highest resignations at 26% (better than KModes outcome)¶

Investigating Remuneration via Kmeans Clustering¶

Nobody in the Remuneration Cluster 0 has resigned, while they do do overtime, they have a higher monthly income than others that do overtime and have resigned¶

Nobody in the Remuneration Cluster 1 has resigned, while they are of a young age category at a mean age of 32 years old, they are distinctly characterised by no Overtime¶

All people in this Remuneration Cluster 3 have resigned. They are mostly on low incomes and are mostly under the age of 35.¶

	Resigned	AverageWeeklyHoursWorked	OverTime	YearsAtCompany	TotalWorkingYears	YearsInRole	Age	YearsWithCurrManager	MaritalStatus	MonthlyIncome	BusinessTravel	BusinessUnit	labels6
0	Yes	63.0	Yes	6	8	4	41	5	Single	5993	Travel_Rarely	Sales	4
1	No	40.0	No	10	10	7	49	7	Married	5130	Travel_Frequently	Consultants	0
2	Yes	50.0	Yes	0	7	0	37	0	Single	2090	Travel_Rarely	Consultants	1
3	No	48.0	Yes	8	8	7	33	0	Married	2909	Travel_Frequently	Consultants	1
4	No	40.0	No	2	6	2	27	2	Married	3468	Travel_Rarely	Consultants	1

		count
labels6	Resigned
0	No	343
0	Yes	45
1	No	361
1	Yes	128
2	No	124
2	Yes	5
3	No	138
3	Yes	26
4	No	196
4	Yes	26
5	No	71
5	Yes	7

	Resigned	OverTime	MaritalStatus	BusinessTravel	BusinessUnit
count	489	489	489	489	489
unique	2	2	3	3	3
top	No	No	Married	Travel_Rarely	Consultants
freq	361	352	209	353	382

	OverTime_Yes	MonthlyIncome	Resigned_Yes	Age	cluster
count	155.0	155.000000	155.0	155.000000	155.0
mean	1.0	5325.722581	0.0	35.961290	0.0
std	0.0	2470.199928	0.0	7.849133	0.0
min	1.0	1129.000000	0.0	19.000000	0.0
25%	1.0	3535.000000	0.0	30.000000	0.0
50%	1.0	4930.000000	0.0	35.000000	0.0
75%	1.0	6563.500000	0.0	41.000000	0.0
max	1.0	13603.000000	0.0	60.000000	0.0

	OverTime_Yes	MonthlyIncome	Resigned_Yes	Age	cluster
count	352.0	352.000000	352.0	352.000000	352.0
mean	0.0	3857.948864	0.0	32.286932	1.0
std	0.0	1523.547554	0.0	5.840721	0.0
min	0.0	1051.000000	0.0	18.000000	1.0
25%	0.0	2550.750000	0.0	28.000000	1.0
50%	0.0	3630.000000	0.0	32.000000	1.0
75%	0.0	4821.500000	0.0	36.000000	1.0
max	0.0	8639.000000	0.0	48.000000	1.0